The frontotemporal EEG as a potential biomarker of early MCI: a case-control study | BMC Psychiatry

0

Speakers

The participants in this study were 120 people (67% women) aged between 40 and 91 years old. The study was conducted from 2016 to 2019. The average age of participants was 67.0 ± 9.19 years. EEG, MRI (specifically, FA-BHQ and GM-BHQ values ​​were obtained), and cognitive screening tests were performed to screen participants. However, only the MMSE is used in this study as a variable. This study was approved by the ethics committees of Keio University and performed in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participants or their legal guardian. This study is a case-control study and is observational in nature; all data obtained was made available to the caregiver and/or participant. The study protocol was approved by the Keio University Ethics Committee with approval numbers: 28–20, 28–59, 29–33, 30–96, 31–56.

The exclusion criteria for patients were: 1) people who have physical or psychiatric disorders that preclude the use of EEG; 2) people with comorbid psychiatric disorders other than dementia; 3) people who have comorbidities that could interfere with EEG recordings, such as brain tumors, strokes, or epilepsy.

For comparison, reference data for healthy individuals that were obtained separately from this study were used. Inclusion criteria for healthy individuals were: 1) no history of mental illness; 2) legal adult defined by Japanese law (age ≥ 20 years old). The age of the healthy volunteer corresponds as closely as possible to that of patients with dementia. It was also required that they did not meet the exclusion criteria for patients with dementia listed above.

All participants were Japanese (Asian) and were divided into three groups: dementia (NOT= 10), MCI (NOT= 33) and command (NOT= 77). Of all participants, 7 were diabetic, 7 were obese, 24 had hyperlipidemia, 4 were diagnosed as clinically depressed, 3 had a history of neurological disease, 35 had hypertension, 2 had a history of stroke, 1 had a history of myocardial infarction, 14 had a history of allergic rhinitis, 1 had a history of COPD, 4 had asthma, 12 had skin problems, 9 had arthritis, 20 had low back pain, 54 had osteoporosis, 4 had a history of kidney problems and 2 had a history of cancer. Dementia patients and MCI patients in this study were all Alzheimer type.

EEG acquisition

A Participants were asked to wear a single channel EEG device (NeuroSky Single Channel EEG, original noise reduction BMD version). The EEG was taken during a relaxed state with eyes closed for a total of 100 s. The electrode location was Fp1 according to the international 10–20 system (left prefrontal region) and the measuring device was MindWave Mobile II BMD II ver. with a sampling rate of 512 Hz. The Mini-Mental State Examination (MMSE) was used as a cognitive screening test. The MMSE is a 30-item question commonly used to assess dementia. Subjects who scored 24 points or less were labeled as demented while patients with a score of 25 to 27 were labeled as MCI. Subjects with a score of 28 or greater are labeled as healthy.

Analysis

Data preprocessing

For the EEG device used in this study, independent verification has already demonstrated that the device can reliably suppress environmental noise and unwanted frequencies. [6]. Signals acquired from Fp1 using a monopole EEG were passed through a 1–30 Hz bandpass filter to extract EEG components [7]. However, even though off-target frequencies can be suppressed, the acquired data is accompanied by noise caused by muscle movement or blinking. To remove these noises, a filter created for this purpose was used. This filter acquires body movement and blinking patterns in advance, and the threshold value is automatically set according to the situation. We adopted conventional methods like noise reduction [6, 8, 9] This procedure reduces computational costs and suppresses blinking, body motion, and electrical noise in real time.

In order to account for individual differences in EEG amplitude, normalization was performed with a mean of 0 and a dispersion of 1 for the filtered signal. Subsequently, a fast Fourier transform was performed to calculate the power spectrum. In addition, to clarify the difference between healthy individuals and each group of patients with dementia and MCI, the mean power spectrum of the comparison target group was set to 1; that is, a relative power spectrum value was used.

The sampling interval of the EEG device was set at 512 Hz. Therefore, the amount of data for each individual was 51,200 samples of 100 × 512. The EEG data of each individual was translated into the domain frequency by Fourier transform per second.

statistical analyzes

First, all acquired data were divided into three groups based on their MMSE score: patients with dementia, patients with MCI, and healthy controls. Descriptive statistics were used to describe study participants. Distributions of all variables were inspected using histograms, qq plots, and Shapiro-Wilks tests before performing statistical analyses. Statistical significance was set to two tails p

Signal processing

As a pre-processing, noise removal was performed on the obtained EEG data and then transformed into the frequency domain. Noise removal was performed using the Summation of Derivatives in Windows (SDW) algorithm and Ensemble Empirical Mode Decomposition (EEMD).

The SDW method detects noise using the sum of the first derivatives in a window [10]. The window selected in this study is 2 s, according to the classical methods, presented in Table 1. The signal was then decomposed into several intrinsic mode functions (IMF) by applying EEMD in the interval detected by the SDW method.

Table 1 Conventional EEG studies using a 2 s window

Components with an average cross-correlation function between MFIs greater than 0.5 were defined as noise and removed. The remaining components were then added to reconstruct the clean signal. For the frequency transformation, the short-term Fourier transform (STFT) was used. The frequency characteristics were the 1-45 Hz frequency slices, averaged across windows, and then normalized. The average of the power spectra for each EEG band was also calculated and used as frequency characteristics. The specifications of the EEG bands were as follows: Delta (δ): 1–4 Hz, Theta (θ) 4–8 Hz, Alpha (α) 8–13 Hz, Alpha-1 (α1) 8–9 Hz, Alpha- 2 (α2) 9–11 Hz, Alpha-3 (α3) 11–13 Hz, Beta (β) 13–30 Hz, Beta-1 (β1) 13–20 Hz, Beta-2 (β2) 20–30 Hz , and Gamma (γ) 30–45 Hz.

Then, the subjects were labeled according to their MMSE score: dementia, MCI and healthy. Since there was an imbalance in the number of samples, the SMOTE (Synthetic Minority Over-sampling Technique) algorithm was applied to artificially create new samples. [30].

In order to verify whether the frequency characteristics used were effective in discriminating between classes, a test of significant difference between classes was carried out. First, a non-parametric test, the Kruska-wallis test, was performed. Then, a multiple comparison with Bonferroni correction was performed. The significance level was set at 5%.

Then, the statistically significant features were used as predictors for support vector machines (SVM) to solve the classification problem. Although SVM is known to be able to handle linearly separable data, using the kernel trick and calculating the maximum headroom, it can handle non-linear data enough.

In this study, SVM with radial basis function kernel was used as a classifier. In order to assess the performance of the classification, a tenfold cross-validation was carried out.

Share.

Comments are closed.